Foot placement seemingly suboptimal

[LouKordos]

Foot placement seems weird, unreliable and unstable. It also seems to keep one foot behind the torso to be able to stabilize, even though the natural walking gait should not be like that

[Add videos showing behaviour]

What has been tried:

• Constrain vel_vector given to foot_pos_desired_world to be between 0 and vel_y_desired, which seems to have improved things a bit.
• Ramping velocity up instead of step
• Waiting for proper contact state before starting to walk, while letting foot position formula and MPC know about future change of velocity. This is necessary because the initial position of the biped in the simulation has both feet near the ground (but still in the air) and the controller instantly wants to start walking. This is still something worth looking into / validating.

Possible solutions:

• Increase vel_y_desired (with and without setting pos_y_desired to actual pos_y at every iteration) to see if it simply does not have enough momentum for putting the foot in front
• Further look into constraining parts of the formula, as it seems placing the foot too far forward makes the robot lose too much velocity, thus breaking.
• Try different combinations of r_x_limit and gait_gain with possible fixes above.
• Waiting for proper contact state before starting to walk, while letting foot position formula and MPC know about future change of velocity. This is necessary because the initial position of the biped in the simulation has both feet near the ground (but still in the air) and the controller instantly wants to start walking. This is still something worth looking into / validating.

[LouKordos]

Apart from tuning, this is not relevant anymore, since turning in place, walking circles and walking at different velocities has successfully been tested for extended periods of time. However, it might be relevant for handling higher desired velocities (>0.6m/s).